如何从嵌套对象数组中获取唯一的搜索结果？

Question

我有一个包含对象数组的文档，该数组具有带有逗号分隔标签的属性。其他文件可以有相同的玩具。

  [
    {
      "documentId": 1,
      "toys": [
        {
          "id": "A",
          "tags": "foo, bar, baz"
        },
        {
          "id": "B",
          "tags": "test"
        },
        {
          "id": "C",
          "tags": "bar"
        }
      ]
    },
    {
      "documentId": 2,
      "toys": [
        {
          "id": "A",
          "tags": "foo, bar, baz"
        },
        {
          "id": "B",
          "tags": "test"
        }
      ]
    }
  ]

我需要基于用户搜索（在标签中）的独特玩具对象列表。这是我想出的，但这对于整个索引来说是独一无二的。不完全是我需要的。

GET my-index/_search
{
  "size": "0",
  "query": {
    "nested": {
      "path": "toys",
      "query": {
        "bool": {
          "must": [
            {
              "match": {
                "toys.tags.ngram": {
                  "analyzer": "standard",
                  "query": "bar",
                  "operator": "AND"
                }
              }
            }
          ],
          "should": [
            {
              "match_phrase": {
                "toys.tags": {
                  "query": "bar"
                }
              }
            }
          ]
        }
      }
    }
  },
  "aggs": {
    "top_matches": {
      "nested": {
        "path": "toys"
      },
      "aggs": {
        "top-result": {
          "top_hits": {
            "size": 1
          }
        }
      }
    }
  }
}

我希望根据“bar”的搜索得到 A 和 C。有什么想法吗？

Answer 1

您可以使用父级和同级

terms

聚合。查询文本位于父级

include

聚合的

terms

参数中。搜索查询结果可能有重复，但聚合查询结果不会有重复

映射

PUT /my-index
{
    "mappings": {
        "properties": {
            "toys": {
                "type": "nested",
                "properties": {
                    "id": {
                        "type": "keyword"
                    },
                    "tags": {
                        "type": "text",
                        "fielddata": true
                    }
                }
            }
        }
    }
}

tags

字段必须具有带有

fielddata

的

true

参数。它需要

terms

聚合

您的文件

PUT /my-index/_bulk
{"create":{"_id":1}}
{"documentId":1,"toys":[{"id":"A","tags":"foo, bar, baz"},{"id":"B","tags":"test"},{"id":"C","tags":"bar"}]}
{"create":{"_id":2}}
{"documentId":2,"toys":[{"id":"A","tags":"foo, bar, baz"},{"id":"B","tags":"test"}]}

聚合查询

GET /my-index/_search?filter_path=aggregations.inside_toys.by_tag
{
    "aggs": {
        "inside_toys": {
            "nested": {
                "path": "toys"
            },
            "aggs": {
                "by_tag": {
                    "terms": {
                        "field": "toys.tags",
                        "include": "bar",
                        "size": 10
                    },
                    "aggs": {
                        "by_id": {
                            "terms": {
                                "field": "toys.id",
                                "size": 10
                            }
                        }
                    }
                }
            }
        }
    }
}

回应

{
  "aggregations" : {
    "inside_toys" : {
      "by_tag" : {
        "doc_count_error_upper_bound" : 0,
        "sum_other_doc_count" : 0,
        "buckets" : [
          {
            "key" : "bar",
            "doc_count" : 3,
            "by_id" : {
              "doc_count_error_upper_bound" : 0,
              "sum_other_doc_count" : 0,
              "buckets" : [
                {
                  "key" : "A",
                  "doc_count" : 2
                },
                {
                  "key" : "C",
                  "doc_count" : 1
                }
              ]
            }
          }
        ]
      }
    }
  }
}